Total Skills
9
Author Profile: majidraza1228
Skills published by majidraza1228 with real stars/downloads and source-aware metadata.
Total Stars
0
Total Downloads
0
Skills Performance
Comparison chart based on real stars and downloads signals from source data.
build-review-interface
0
Stars
0
Downloads
0
error-analysis
0
Stars
0
Downloads
0
eval-audit
0
Stars
0
Downloads
0
eval-coding-agent
0
Stars
0
Downloads
0
eval-tool-use
0
Stars
0
Downloads
0
evaluate-rag
0
Stars
0
Downloads
0
generate-synthetic-data
0
Stars
0
Downloads
0
validate-evaluator
0
Stars
0
Downloads
0
Published Skills
build-review-interface
Build a custom browser-based annotation interface for reviewing LLM traces and collecting human labels. Use when reviewers are working with raw JSON files, when you need to collect Pass/Fail labels at scale, or when trace data needs domain-specific formatting to be readable.
Repository SourceNeeds Review
error-analysis
Systematically identify and categorize failure modes in an LLM pipeline by reading traces. Use when starting a new eval project, after significant pipeline changes (new features, model switches, prompt rewrites), when production metrics drop, or after incidents.
Repository SourceNeeds Review
eval-audit
Audit an LLM eval pipeline and surface problems: missing error analysis, unvalidated judges, vanity metrics, etc. Use when inheriting an eval system, when unsure whether evals are trustworthy, or as a starting point when no eval infrastructure exists.
Repository SourceNeeds Review
eval-coding-agent
Evaluate a coding agent's output quality across the failure modes specific to code generation and editing: correctness, scope discipline, instruction following, safety, and diff quality. Use when building or improving a Claude-powered coding assistant, code review agent, or code generation pipeline.
Repository SourceNeeds Review
eval-tool-use
Evaluate whether an LLM agent selects the right tools, constructs correct arguments, sequences tool calls appropriately, and handles errors gracefully. Use for any Claude agent that has access to tools (function calling, MCP servers, API integrations).
Repository SourceNeeds Review
evaluate-rag
Evaluate a RAG (retrieval-augmented generation) pipeline's retrieval quality and generation quality separately. Use when the pipeline retrieves context from a knowledge base before generating answers.
Repository SourceNeeds Review
generate-synthetic-data
Create diverse synthetic test inputs for LLM pipeline evaluation using dimension-based tuple generation. Use when bootstrapping an eval dataset, when real user data is sparse, or when stress-testing specific failure hypotheses. Do NOT use when you already have 100+ representative real traces.
Repository SourceNeeds Review
validate-evaluator
Calibrate an LLM judge against human labels using data splits, TPR/TNR, and bias correction. Use after writing a judge prompt (write-judge-prompt) to verify alignment before trusting its outputs in production.
Repository SourceNeeds Review
write-judge-prompt
Design LLM-as-Judge evaluators for subjective criteria that code-based checks cannot handle. Use when a failure mode requires interpretation (tone, faithfulness, relevance, completeness, reasoning quality). Do NOT use when the failure mode can be checked with code (regex, schema validation, test execution). Default to Claude (claude-sonnet-4-6 or claude-opus-4-6) as the judge model.
Repository SourceNeeds Review